2.2 Potential explanatory measures
2.2.1 Cortical architecture
were available for area degree than for the projection-based connectivity measures.
Parts of this section have been published in Beul et al. (2015), Beul et al. (2017) and Beul and Hilgetag (2019a).
2.2. Potential explanatory measures
PL
type n.a.
1 2 3 4 5 lateral medial
6m
20b DLS
21b PMLS
PLLS 19
17 1918 21a ALLS
AMLS
SSF 7
AES SIV SSAi
SSAo
DP
V VP P 5bm
5al5bl 5am
POA PFCdl
4
PFCr 6l
4g 3a 3b
1
2 SII
36 35
20a
PS Ia
Ig AAF
EPp
AII AI
Tem PFCv
VLS
PFCr SVA
LA 5m
3a3b 4 6m
PFCv PFCdm 1 CGa
CGp
20a 20b
RS
187
18 17
36 35 ER
PSb
OB
IL PSb PL
Figure 2.1:Architectonic type in the cat cortex. Parcellation of the cat cortex, adapted from Scannell and colleagues (1995). Areas were assigned to architectonic types 1–5 according to their level of architectonic differentiation.Type n.a.: no architectonic type was assigned.
Abbreviations as in Supplementary Table D.1.
complex cortical structure into a single parameter (e.g. Barbas, 1986; Barbas and Rempel-Clower, 1997; Rempel-Clower and Barbas, 2000; Barbas et al., 2005; Hilgetag and Grant, 2010). Thereby, areas are categorised along a spectrum of architectonic types, ranging from poorly differentiated types, with low neuron densities and few layers that are hard to demarcate, to highly differentiated types, with numerous, clearly distinguishable layers and high neuron densities.
Cortical areas were rated on an ordinal scale based on several criteria for their architectonic differentiation, assigning an architectonic type to each area. One major feature was the relative width, density and granularisation of layer 4 (cf. Barbas, 1986). Our classification thus follows the classical tradition of using cytoarchitectonic features for characterising cortical areas as practised since the early 20th century (Brodmann, 1909; von Economo, 1927).
In the macaque cortex, the rating was performed by Helen Barbas, assigning types ranging from 1 (least differentiated) to 8 (most differentiated). These architectonic types for the adult macaque cortex have been published previously for the cortical areas that we considered in our analyses of connectivity in the immature macaque cortex (Hilgetag et al., 2016). In the cat cortex, this rating was performed by Simon Grant, assigning types ranging from 1 (least differentiated) to 5 (most differentiated).
In the ranking procedure, first, areas of highest and lowest architectonic differenti-ation were identified and assigned to the architectonic types 5 and 1, respectively.
Second, areas in which cortical layers could be distinguished almost as well or as badly as in areas of types 5 and 1 were assigned the architectonic types 4 and 2, respectively. All remaining areas, necessarily of an intermediate differentiation, were assigned to architectonic type 3. For photographic examples of architectonic types
see Hilgetag and Grant (2010). Using these criteria, 49 areas across the whole cat cortex were ranked. Figure 2.1 depicts the assigned architectonic types in the cortical parcellation of Scannell and colleagues (1995). From this architectonic type ranking, we determined the difference between the architectonic types, ∆type (cf.
Barbas, 1986), of any two of the cortical areas with a defined architectonic type, where∆type =typesource area−typetarget area.
Neuron density A quantitative measure that reflects architectonic differentiation is overall neuron density (Dombrowski et al., 2001). We used an unbiased quantitative stereologic approach to estimate neuron density in the macaque cortex from coronal sections that were stained to mark neurons using either Nissl stain or immunohisto-chemical staining for neuronal nuclei-specific antibody (NeuN), which labels neurons but not glia, using a microscope-computer interface (StereoInvestigator, MicroBright-Field Inc., Williston, VT). We verified that there was a close correspondence between measures derived from both staining methods in a sample of areas for which both measures were available (r= 0.99,p= 0.001), and accordingly transformed density measures from different staining methods to a common reference frame. The neuron density measurements used here have partly been published previously (Dombrowski et al., 2001; Hilgetag et al., 2016). In total, neuron density measures were available for 48 of the 91 areas of the M132 parcellation (Figure 2.2). Within the 29×29 subgraph of areas injected with retrograde tracer, neuron densities were available for 14 of the 17 core areas and 10 of the 12 non-core areas.
We quantified how similar areas were in their neuron density by computing the log-ratio of neuron density values for each pair of areas (which is equivalent to the difference of the logarithms of the area densities). Specifically, loд-ratiodensity = ln(densitysource area/densitytarget area). This procedure enabled us to directly relate each sampled projection to the density ratio of its source and target area. The use of a logarithmic scale was indicated, since the most extreme value of the neuron density measures was more than three standard deviations above the mean of the considered neuron densities (Buzsáki and Mizuseki, 2014). For analyses which required considering an undirected equivalent of the actual neuron density ratio, we used the absolute value of the log-ratio, |log-ratiodensity|. To relate the neuron density ratio to ranked projections strength, we also ranked the absolute value of the log-ratio of neuron density, |log-ratiodensity|, separately per target area. That is, the smallest absolute neuron density ratio was ranked highest (as rank 1) for each injected area, and successively larger absolute neuron density ratios were ranked accordingly (increasing rank number). Hence, areas of similar neuron density (small absolute ratio) were ranked higher than areas of strongly diverging neuron density
2.2. Potential explanatory measures
V1
V2 V3
V4 Prostriate
31 23 7m V6A
V6 MIP
V3A
MT V4t
TEO TEOm
Tpt 1
PIP
TEa/m pIPa
F7 F4 F1 F2
3
PGa 7A MST
FST STPr STPi STPc
7op 7B AIP
2 5 LIP VIP
DP 44 8l45b 9/46v
9/46d8r 46d 46v 8m
8B 9 10 14
TH/TF TEav
TEpv TEad TEpd TEa/m a
OPRO
Temp. Pole
Perirhinal Entorhinal Subiculum
PBr PBc
SII 45a
OPAl
piriform 25 32 24d
F6 24c F3
MB
LBcore Para Ins
ProMGu F5
11 13
11
12 29/30
24a 24b
L3 dendritic tree size [µm²]
> 30 000
> 40 000
> 50 000
> 60 000
> 70 000
> 80 000
> 90 000
> 100 000
> 110 000
> 120 000
> 130 000
> 180 000
L3 dendritic tree size
L3 spine density [#/10 µm]
>= 6
>= 8
>= 12
>= 14
>= 16
>= 18
>= 20
>= 22
>= 24
>= 26
>= 28
>= 30
>= 32 V1
V2 V3
V4 Prostriate
31 23 7m V6A
V6 MIP
V3A
MT V4t
TEO TEOm
Tpt 1
PIP
TEa/m pIPa
F7 F4 F1 F2
3
PGa 7A MST
FST STPr STPi STPc
7op 7B AIP
2 5 LIP VIP
DP 44 8l45b 9/46v
9/46d8r 46d 46v 8m
8B 9 10
14
TH/TF TEav
TEpv TEad TEpd TEa/m a
OPRO
Temp. Pole
Perirhinal Entorhinal Subiculum
PBr PBc
SII 45a
OPAl
piriform 25 32 24d
F6 24c F3
MB
LBcore Para Ins
ProMGu F5
11 13
11
12 29/30
24a 24b
L3 spine density
L3 spine count [#]
> 500
> 1 000
> 2 000
> 3 000
> 4 000
> 6 000
> 7 000
> 8 000
> 9 000
> 10 000 V1
V2 V3
V4 Prostriate
31 23 7m V6A
V6 MIP
V3A
MT V4t
TEO TEOm
Tpt 1
PIP
TEa/m pIPa
F7 F4 F1 F2
3
PGa 7A MST
FST STPr STPi STPc
7op 7B AIP
2 5 LIP VIP
DP 44 8l45b 9/46v
8r 9/46d46d
46v 8m
8B 9 10 14
TH/TF TEav
TEpv TEad TEpd TEa/m a
OPRO
Temp. Pole
Perirhinal Entorhinal Subiculum
PBr PBc
SII 45a
OPAl
piriform 25 32 24d
F6 24c F3
MB
LBcore Para Ins
ProMGu F5
11 13
11
12 29/30
24a 24b
L3 spine count
L3 soma size [µm²]
> 120
> 130
> 180
> 210
> 220
> 230
> 240
> 260
> 280
> 290
> 310
> 330 V1
V2 V3
V4 Prostriate
31 23 7m V6A
V6 MIP
V3A
MT V4t
TEO TEOm
Tpt 1
PIP
TEa/m pIPa
F7 F4 F1 F2
3
PGa 7A MST
FST STPr STPi STPc
7op 7B AIP
2 5 LIP VIP
DP 44 8l45b 9/46v
8r 9/46d46d
46v 8m
8B 9 10
14
TH/TF TEav
TEpv TEad TEpd TEa/m a
OPRO
Temp. Pole
Perirhinal Entorhinal Subiculum
PBr PBc
SII 45a
OPAl
piriform 25 32 24d
F6 24c F3
MB
LBcore Para Ins
ProMGu F5
11 13
11
12 29/30
24a 24b
L3 soma size
neuron density [neurons/mm³]
> 35 000
> 40 000
> 42 500
> 45 000
> 47 500
> 50 000
> 52 500
> 55 000
> 60 000
> 65 000
> 75 000
> 80 000
> 85 000
> 100 000
> 150 000 V1
V2 V3
V4 Prostriate
31 23 7m V6V6A
MIP
V3A
MT V4t
TEO TEOm
Tpt 1
PIP
TEa/m pIPa
F7 F4 F1 F2
3
PGa 7A MST
FST STPr STPi STPc
7op 7B AIP
2 5 LIP VIP
DP 44 8l45b 9/46v
8r 9/46d46d 8m 46v
8B 9 10 14
TH/TF TEav
TEpv TEad TEpd TEa/m a
OPRO
Temp. Pole
Perirhinal Entorhinal Subiculum
PBr PBc
SII 45a
OPAl
piriform 25 32 24d
F6 24c F3
MB
LBcore Para Ins
ProMGu F5
11 13
11
12 29/30
24a 24b
neuron density
Figure 2.2:Variation of cytoarchitectonic features across the macaque cortex.
Figure 2.2:Variation of cytoarchitectonic features across the macaque cortex. Variation of neuron density, L3 neuron soma cross section, L3 dendritic spine count, L3 dendritic spine density and L3 dendritic tree size depicted on the M132 parcellation (Markov et al., 2014a).
For grey areas, no values were available. See Supplementary Table D.4 for correspondences between areas in the M132 parcellation and alternative parcellations. Abbreviations as in Markov et al. (2014a).
(large absolute ratio) relative to each of the injected areas.
From the available neuron density measures we were able to determine the relative architectonic profile for 1128 of the sampled projections. After applying a threshold of constituent neurons to exclude potentially unreliableNSG% values (discarding projections comprising less than 20 neurons), this included 172 projections with an associatedNSG% for the subset of projection published in Markov and colleagues (2014b) and 521 projections with an associatedNSG% for the complete set of injec-tions published in Chaudhuri and colleagues (2015).
Cellular morphological measures
Measures of cellular morphology characterise individual cells, and thus provide an impression of an area’s constituting elements, but not of its overall architectonic differentiation. The measures of cellular morphology we considered were mostly reported by Elston and colleagues (Elston and Rosa, 1997, 1998a,b; Elston et al., 1999a,b; Elston, 2000; Elston et al., 2001; Elston and Rockland, 2002; Elston et al., 2005, 2009, 2010a,b, 2011a,b; Coskren et al., 2015; Gilman et al., 2017). Specifically, four aspects of L3 pyramidal neuron morphology were measured across the macaque cortex: the cross section of the cell soma (soma cross section), the average total spine count on the basal dendritic tree (spine count), the peak density of dendritic spines (spine density), and the size of the basal dendritic tree (dendritic tree size).
Spine density was measured as the number of spines per 10 µm dendrite segment, and peak spine density was then calculated as the average density along the five consecutive 10 µm segments that yielded the highest spine density (see e.g. Elston and Rosa, 1998b). Supplementary Table D.4 gives an overview of the correspondence between the parcellations used in the morphological data references and the M132 parcellation, as well as the relevant reports. Specifically, in the M132 parcellation, soma cross section was available for 30 areas, spine count and spine density for 33 areas, and dendritic tree size for 34 areas (Figure 2.2). The soma cross section may be related to the overall size of a neuron, which would be characterised by
2.2. Potential explanatory measures
further properties such as soma surface area and soma volume. However, given the varying shapes of somata, inferences from the cross-sectional area to overall soma size are not straightforward. Such inferences are further impaired by the difficulty of measuring the cross section at comparable locations across different neurons. For example, some measures of cross-sectional area that we included in our analyses were taken at the level of the basal dendritic tree (e.g. Elston et al., 2011b), while others were taken at the widest point of the cell body (e.g. Elston and Rosa, 1997;
Gilman et al., 2017). To quantify how similar areas were in the four morphological measures across the cortex, we computed the difference of their values for each pair of areas, where∆morphological measure =measuresource area−measuretarget area. This resulted in∆soma cross section,∆spine count,∆spine density, and∆tree size. Each of these dif-ference measures was converted to an undirected variable by computing its absolute value, |∆soma cross section|, |∆spine count|, |∆spine density|, and |∆tree size|, where appro-priate. To relate the four morphological measures to ranked projections strength, we ranked their absolute difference measures separately per target area, analogous to the ranking described for the absolute neuron density ratio above. That is, smaller absolute difference measures were ranked highest (rank 1), and successively larger absolute difference measures were ranked accordingly (increasing rank number).